Overview
Speak one language and be heard in another, in real time, with the Voice Translation SDK or the Krisp Voice Translation API.
Voice Translation turns speech in one language into speech in another, in real time. Speech is transcribed, translated and synthesized in the target language, and you can also receive the original and translated transcripts.
Choose SDK or API
| Voice Translation SDK | Voice Translation API | |
|---|---|---|
| What it is | A client SDK that captures audio and streams it to Krisp's Voice Translation service | A WebSocket endpoint: wss://streaming.krisp.ai/vt |
| Get started | SDK quick start | Voice Translation API reference |
| Authentication | Your Krisp API key or a Voice Translation session token, see the SDK quick start | A session token with usage_type: vt |
| Audio | Captured by the SDK | pcm_s16le or opus, 16 kHz, mono |
Key points
- Each session is single-directional: one source language to one target language. For a two-way conversation, open two sessions, one per direction.
- Translation is all-to-all across 65+ supported languages, using BCP-47 language codes.
- Latency: after a short conversational delay of several words, steady-state latency is typically 800–900 ms.
- Before going live, read Best practices and Rate limiting.
Updated about 13 hours ago
Did this page help you?